In February 2012, after 10 years of research and the acquisition of a supercomputer, proof is given that Oware -the oldest game of the world- can not the solved. While opening new research avenues, this research provides insights on the solution found in 2002 on a variant created by computer scientists.
Speaker at the Board Games Studies Colloquium XV in Haar, Germany, April 20, 2012.
Oware: The Oldest Game of the World Will Not Be Solved by Computers
1. Oware: The Oldest Game of the World
Will Not Be Solved by Computers
Xavier A. P. Blanvillain
Oware European Vice-Champion 2000, Elo 2048, Amsterdam, The Netherlands
Co-founder of the International Association of Warri Players, Switzerland
Member of the Oware Society, United Kingdom
xblanvillain@smartproduction.fr
Category: computer science, theory
Keywords: oware, warri, awari, depth-first search,
backward induction, retrograde analysis, endgame book,
supercomputing.
On September 24th, 2011, all of Oware’s positions
Abstract
were searched with a single computer and a mono-
threaded brute-force algorithm. It took 12 days, 1 hour
The 5,000 year-old game Oware, nowadays often
and 22 minutes to report a space-state of 827,240,309,058
referred to as the ‘Chess of the Caribbean’ or Warri, is one
legal positions. While analyzing the gap with the results
of the five classical strategic board games used in game
from Awari’s solution (reporting the upper bound of
theory, together with Chess, Checkers, Othello and Go.
889,063,398,406 positions), remarkable patterns for illegal
The capabilities of computers in the 1990’s were limited to
positions were found, positions that can never arise in the
small-scale problems when applied to game theories and
course of a game. As a side note, during the search, a loop
solving algorithms. The classical strategic board games
with all 48 seeds was revealed as well as a game where 2
were too demanding of computer resources to allow a
players can play more than 7.9 billion moves and yet keep
resolution in an acceptable timeframe. The less complex
all the 48 seeds.
family of mancala games (to which Oware belongs), was
therefore the ideal candidate for applying and validating
Under Abapa rules, loops are the key constraint in
these algorithms. In 1991, the ‘Awari’ variant was created
solving Oware. There is no rule that a computer can apply
and became the reference for use by computer scientists.
to solve these loops. Although patterns have been
It was used not only in computer science research but also
identified, the loops need to be solved on a case-by-case
in computer programming competitions organized by the
basis by humans. The use of backward induction to solve
Mind Sports Olympiad. In 2002, a team of computer
the game, starting from the endgames, helps in reducing
scientists from The Netherlands found a solution to this
the number of loops. Only loops with 10 seeds or less are
variant by applying a parallel retrograde analysis using a
commonly encountered by players, and these have been
network of computers.
collected in an endgame book.
Despite the major accomplishment of solving Awari,
On February 15th, 2012, results lead to the conclusion
this variant deviates from the international ‘Abapa’ rules
that Oware will not be solved by a computer without
followed by Oware players and its organizations: the
human intervention: the solution for loops with 2, 3, 4, 7
Oware Society and the International Association of Warri
and 9 seeds had to be entered manually into the database.
Players. Awari, unlike Oware, uses the ‘Grand slam’ rule
Furthermore, an 11-seed loop and a 13-seed loop were
and creates an arbitrary rule to solve the rotating positions
found, which cannot be solved with Abapa rules. These
(loops). This was key to solving Awari. To date, the
loops have blocked the resolution of the database which
official game Oware has not been solved.
raises the question whether Oware is solvable at all.
1
2. Biography
Xavier Blanvillain, born in France
in 1969, has been playing Oware
for over 20 years. He first started
competing in 1999, in competitions
mainly organized by the Mind
Sports Olympiad, the Oware
Society and the International
Association of Warri Players, which
he co-founded in 2008 in
Switzerland. He has won prizes at
both national and international
levels, including Belgian champion in 1999, Italian
champion in 2000, UK champion in 2002 and European
vice-champion in 2000. Solving the game of Oware has
been his passion for the last 10 years. He acquired a
supercomputer that made this research possible in 2011.
Xavier holds a MBA from the Rotterdam School of
Management and a MSc from the French Grande École
ESEO. He has worked as an entrepreneur and a
consultant, and currently works in upper management in
the Consumer Electronics industry.
2